FPS COMPARISON 1.02.14821 vs 1.03.15527 @ 09-08-2011

[SPUDLEY1977]

Hi all, based on recorded FPS over the Black Death Track here are the graphs.

1920 x 1080
Q9550 GTX 570 Win7 64 4G DDR2
no OC'ing, all stock.

Both tracks were synced up over a total of 226 seconds of recorded FPS during the Black Death. Over this period the AVERAGE FPS prepatch was 31 and post beta it was 24 for a delta of 7.

Particularly interesting is that for the first 1/2 of the time the FPS hit between 5 and 10 FPS, but over the remainder there are large sections where for extended periods the decrease in FPS swings in large blocks from 10 to 20 and 30 FPS.

This pattern in many regards matched online MultiPlayer. In game the new Beta is noticeably less smooth, and based on our comparisons last night we noticed the new: sounds, explosive effects, tracer, beaches, water color and waves, sun blinding, loss of clicks in cockpits, the more subtle in cockpit shadows, excessive fog in all depths of atmosphere, fog painted over all ground textures ( like spilled milk) even at close up.
Terrain textures appear a bit hazed out / blurry.

Both versions have components that beg attention. But overall the FPS drop IS noticeable, the fog is way too permissive and painted on EVERYTHING, the sound is a large step in the right direction but it needs proper Mixing and has some obvious shortcomings.

So there you have it an objective analysis of the FPS impact. I have others but now I just want to fly after a long stressful day at the Programmer Ivory Towers (The PIT).

:)

[baronWastelan]

Thanks for posting this, I have a very similar system. I now see the benefit of my upgrading to a Q9650 and OC to 3600mhz! I'm getting 31fps avg @ 1920x1200 with the new beta. Also GTX 570, stock clocks.

[SPUDLEY1977]

baron, you're welcome. However watching CPU and GPU utilization it seems the GPU may be approaching more of a limiting factor, not because it is fully utilized but it appears more % utilized vs the CPU's. Based on Utilization of both it FEELS like it has a bit more to do with the way the code is written like it needs some optimizations, and how data/textures are culled from the HD/ RAM/etc . I fly on the RAVENS Nest so if you are there give me a shout before you send me your BB's.

[baronWastelan]

How's your memory bandwidth? That should be a factor in GPU utilization. I benchmark mine with this free utility:

MaxxMEM² - PreView (http://www.maxxpi.net/pages/downloads/maxxmemsup2---preview.php)

I have 4Gig DDR3 @ 1600mhz

[Icebear]

Very interesting evaluations ! Thank you very much.

@SPUDLEY1977

Wich software did you use ? Have you also compared the VRAM workload ?

[mazex]

An interesting thing regarding CoD performance is that it seems more CPU limited than I had thought, at least in my own empirical test I did a while ago when upgrading my rig... Below is a copy from an old post (from May so two patches away?) :



Below is a rather interesting study of what affects CoD performance. I used my old rig, my old rig with my new GPU and finally my new rig...

The "rules":

* CoD settings to "default" high settings with no AA, SSAO or anti epilepsy.
* Start the Fraps benchmark as soon as the track has started, and stop it the second the track is done. Don't forget to turn off "Stop benchmark after 60 seconds"!
* I used 1680x1050 resolution

On my old system I got this:

Core2Duo E8400@3.3 - GTX275 - 6 GB DDR2 800Mhz - P45Mb

Frames, Time (ms), Min, Max, Avg
6851, 218058, 9, 69, 31.418


Then my OLD rig with my NEW GPU - only gained 1.9 fps in average! Sure CPU limited - but the weird thing is that both cores are never above 70% load, most of the time at ~30-60%

Core2Duo E8400@3.3 - GTX580 - 6 GB DDR2 800Mhz - P45Mb

Frames, Time (ms), Min, Max, Avg
7230, 217293, 13, 70, 33.273


And then - my new rig in full glory (same GFX card as above but massive fps gain):

i7 2600k @ 4.5 - GTX580 - 8 GB DDR3 1600Mhz - P67 Mb

Frames, Time (ms), Min, Max, Avg
14867, 216997, 22, 168, 68.512

All on Windows 7 x64

[Tavingon]

My game runs smooth, but when I shoot up enemy planes and they start to shatter and burst into flames the game lags like hell

[SPUDLEY1977]

Hi Mazex, over several diff PC builds and related evaluations I concluded that with a duo core CPU wherein even when it did NOT show full CPU utilization a simple drop in of a quad core resulted in significant FPS increases and noticeable smoother gameplay for the IL-2 series.

Based on your data, your CPU was limiting despite the fact that you were not showing 90-100 core utilization. I suspect it has something to do with the architecture of the old DUO Cores. The 60% utilizations with spikes into the 70's apparently was in fact a bottleneck. I have experienced this on various PC builds for my personal use as well as paying customers.

Stepping from 275 to the 580/570 is also a huge step in throughput capacity - assuming you are not limitd by a duo core CPU. For example, if you dropped your 275 in your current rig, I would bet you a NEW SET OF TEETH that your FPS, fluidity, and pleasure will drop through the toilet. THEN you would be calling me to unclog with some new PC giblets. Try it please sir - just to confirm as many others are likely to find the results/conclusion useful, and that is how our community adds value to these forum posting/rants. :)

As a side note I bet this current version is not really the new engine they have been working on, it plays like the old engine with modified textures and other bits. Even some of the odd interactions between objects can be seen in CLOD as they were in the groundbreaking Orig Series.

[SPUDLEY1977]

Icedbeer, You're welcome sir always a pleasure to help when appreciated. I did not do the VRAM as it is a bit of a pain in the dierrerrierrerre.

The software used was MSI Afterburner to record the datapoints - kudos to MSI for providing such an excellent piece of software - none compare. I used Excel to arrange and graph the datapoints for your visual pleasure.

[SPUDLEY1977]

Baron von Pizza (I'm hungary), I am not familiar with that, what specificially is it supposed to test, memory bandwith? and which one as there are several memory bandwith potential bottlenecks on a PC....Mobo<>CPU, MOBO >< GPU, RAM bandwidth, and like others that I am not versed in.

Damn Im hungry and it is 1600 hours Fri AM and I cannot sleep so Im browsing,.... make mine extra sauce, pepperoni, veggies, and Jalapenos please. :)

[baronWastelan]

The test utility runs on your CPU to calculate the maximum theoretical throughput of your system RAM. The results are determined by your memory sub-system hardware and how it is configured. It does not test graphics card. Bigger numbers are better for bandwidth, smaller number is better for latency. I provided my results so you can compare yours since you have a similar system, i.e. LGA775 socket quad-core processor and 4 GB RAM.

[SPUDLEY1977]

Baroni,
Thanks but I am all tested out, don't have the software and do not feel the need to this test aspect. I was only providing info for others to assismilate and use for good purpose.

[Icebear]

Icedbeer, You're welcome sir always a pleasure to help when appreciated. I did not do the VRAM as it is a bit of a pain in the dierrerrierrerre.

The software used was MSI Afterburner to record the datapoints - kudos to MSI for providing such an excellent piece of software - none compare. I used Excel to arrange and graph the datapoints for your visual pleasure.

Too bad as IMO the stutters many people suffer could also be caused by excessive VRAM workload or deficient VRAM cache management. But a pain in the dierrerrierrerre is even more worse.....:mrgreen:

[SPUDLEY1977]

Ice,
I have been shopping for an SSD 64GB Crucial M4, as the latest SSD technology is not quite ready for prime time as our HD technology is. Once I locate one at the correct price point I will get one.

[Ze-Jamz]

My game runs smooth, but when I shoot up enemy planes and they start to shatter and burst into flames the game lags like hell

+1

Any types of explosions now i get this lag and massive FPS hit....NOT GOOD

Blimey that with the Spit2 on nearly every server is starting to ruin my game, also the more I play this patch the more that Sun haze is starting to annoy me too..maybe just a tad too bright?

[mazex]

Stepping from 275 to the 580/570 is also a huge step in throughput capacity - assuming you are not limitd by a duo core CPU. For example, if you dropped your 275 in your current rig, I would bet you a NEW SET OF TEETH that your FPS, fluidity, and pleasure will drop through the toilet. THEN you would be calling me to unclog with some new PC giblets. Try it please sir - just to confirm as many others are likely to find the results/conclusion useful, and that is how our community adds value to these forum posting/rants. :)



So true, testing the new rig with my old GPU is naturally what is missing from my tests, the problem was that at that stage my oldest son had inherited the old Core 2 duo / GTX275 rig and he was not interested in me ripping the GPU for some tests, and another patch was out by then so I would have to redo the previous three tests too :) I am very sure that the old GTX275 in the new rig would get seriously bottle necked by the GPU, but I was a bit surprised that the old rig benefitted almost nothing from the new GTX580... As you say I'm pretty sure that the old Core 2 duo architecture is the problem. I had a LOT of problems in RoF with the C2duo with bad stuttering (the problem only seems to happen in Windows 7 x64 though, it worked fine in XP32)...

Will have to do the test....

[Icebear]

Ice,
I have been shopping for an SSD 64GB Crucial M4, as the latest SSD technology is not quite ready for prime time as our HD technology is. Once I locate one at the correct price point I will get one.

hmmm...don't think you got me right. I'm talking about the video RAM of the graphic adapter, not about SSD or HDD. Anyway, I'm running a 256GB Crucial m4 and a 1TB Western Digital Black Caviar at SATA6. Tested CloD on both drives with no remarkable difference, except the loading times.

[LoBiSoMeM]

The "new" graphics engine is just this FPS killer blue filter?

;)

Can't resist... And i'm a fanboy...

[SPUDLEY1977]

Now steam is jacked and the normal connection steps are rejected. Steam blows

[Fall_Pink?]

I noticed there a extra dll's loaded now with this patch. What about if you disable the extra dll's?

<value>...parts\core\fmodex.dll;parts\core\fmod_event.dll</value>

Any difference?

Rgs,
FP

[mazex]

hmmm...don't think you got me right. I'm talking about the video RAM of the graphic adapter, not about SSD or HDD. Anyway, I'm running a 256GB Crucial m4 and a 1TB Western Digital Black Caviar at SATA6. Tested CloD on both drives with no remarkable difference, except the loading times.

A game that gets better performance from a faster disk accept occasional loading times is in serious need of optimization... Regarding RAM memory speed there is in my experience very little performance to gain from that for gaming. A few percent maybe but never in the same league as changing GPU (and the memory bandwidth of that GPU) or CPU (which naturally in many cases gives a good memory speed boost as new CPU:s have faster memory access through new chipsets, memory types etc). But if you have an i7 2600k for example and go from memory clocked at 1600Mhz to 1866Mhz or even 2000Mhz the performance gain in a game like CoD should be hard to measure... Someone sure have some benchmark available.

EDIT:

Link to comparison of different memory speeds and it's impact on the Sandy Bridge chipset (which most of us with new rigs use here):

http://techreport.com/articles.x/20377/3

[Vengeanze]

This is my 3DMark Score before and after adding more RAM:


P12 688 with 4GB RAM
P12 718 with 12GB RAM

[LoBiSoMeM]

The bottlenecks of this sim are CPU+RAM.

ANY graphics settings here have minor FPS impact than number of AI aircrafts and ground units ingame - even if I can't see these units.

I believe that the direction needed is focus in more RAM usage and multicore suport. Looking at the graphics engine only is poor. Test for yourselves.

[Mr Greezy]

A game that gets better performance from a faster disk accept occasional loading times is in serious need of optimization...


Exactly. You can't have a game with massive stutters, an inability to run SLI, and crazy FPS dips and then have "the fix" be a SSD. I've got one...guess what, it still runs like something is broken.

[ACE-OF-ACES]

Hi all, based on recorded FPS over the Black Death Track here are the graphs.

1920 x 1080
Q9550 GTX 570 Win7 64 4G DDR2
no OC'ing, all stock.

Both tracks were synced up over a total of 226 seconds of recorded FPS during the Black Death. Over this period the AVERAGE FPS prepatch was 31 and post beta it was 24 for a delta of 7.

Particularly interesting is that for the first 1/2 of the time the FPS hit between 5 and 10 FPS, but over the remainder there are large sections where for extended periods the decrease in FPS swings in large blocks from 10 to 20 and 30 FPS.

This pattern in many regards matched online MultiPlayer. In game the new Beta is noticeably less smooth, and based on our comparisons last night we noticed the new: sounds, explosive effects, tracer, beaches, water color and waves, sun blinding, loss of clicks in cockpits, the more subtle in cockpit shadows, excessive fog in all depths of atmosphere, fog painted over all ground textures ( like spilled milk) even at close up.
Terrain textures appear a bit hazed out / blurry.

Both versions have components that beg attention. But overall the FPS drop IS noticeable, the fog is way too permissive and painted on EVERYTHING, the sound is a large step in the right direction but it needs proper Mixing and has some obvious shortcomings.

So there you have it an objective analysis of the FPS impact. I have others but now I just want to fly after a long stressful day at the Programmer Ivory Towers (The PIT).

:)Nice Work!

I wish all FPS video testing presented here was done this well!

[Icebear]

hmmm...don't think you got me right. I'm talking about the video RAM of the graphic adapter....

....they don't want to understand me. :( Or do they ignore me ? ;)

[Fall_Pink?]

The bottlenecks of this sim are CPU+RAM.

ANY graphics settings here have minor FPS impact than number of AI aircrafts and ground units ingame - even if I can't see these units.

I believe that the direction needed is focus in more RAM usage and multicore suport. Looking at the graphics engine only is poor. Test for yourselves.

Sorry, have to disagree. I use a i2600 and 8 Gb ram and CoD hardly uses it. 50% of first core is used and only 15 to 20% (often less) of the other ones. ~ 6 Gb ram is enough for this game, but as always the more the better but that's purely meant for Windows to be used.

The bottleneck is still the game itself; it simply does not make optimal use of latest multi core cpu's.

What it does however is, is put a hefty load on the GPU when you run near max video settings.

Rgs,
FP

[LoBiSoMeM]

Sorry, have to disagree. I use a i2600 and 8 Gb ram and CoD hardly uses it. 50% of first core is used and only 15 to 20% (often less) of the other ones. ~ 6 Gb ram is enough for this game, but as always the more the better but that's purely meant for Windows to be used.

The bottleneck is still the game itself; it simply does not make optimal use of latest multi core cpu's.

What it does however is, is put a hefty load on the GPU when you run near max video settings.

Rgs,
FP

So, you are talking the same: CloD don't use multicore CPU power in an optimal way. And the game itself uses only 1.5GB. It's not good if FPS drop with lots of objects NOT IN SIGHT.

CloD uses all GPU power available. And if the game doen not make optimal use of latest mulit core GPU's, the bottleneck IS CPU and RAM...

[SsSsSsSsSnake]

Mazex same system specs as you and since upgrading to those runs really nice,

[mazex]

Mazex same system specs as you and since upgrading to those runs really nice,

I agree, I really don't have any performance issues in CoD. Occasionally it can have some stuttering when something goes boom or some loading occurs - but that can happen in other games too. Even though it feels like it will be possible to get higher fps when the game is optimized it's really OK now on my rig at least. Around 50-70 normally in a battle at low altitude running on "high" settings... I think I get about the same fps in RoF which has a less ground details and objects.